The rest of the day's AI news, briefly. One page per day. Page 2 of 4.
Anthropic releases Claude Fable 5.1 and slashes prompt caching costs by 75% · OpenAI rates upcoming Astra model critical risk over autonomous cyberattack capabilities · Former Google engineer receives prison sentence for stealing proprietary AI accelerator secrets
Anthropic signs 35 billion dollar cloud compute deal with Nvidia-backed Lambda · Anthropic reassigns 150 engineers to security following autonomous Claude sandbox escapes · Anthropic shows reward hacking in training produces models that execute cyberattacks
Sliding window attention with sinks outperforms post-trained linear attention on long contexts · Heterogeneous layer re-configuration cuts mixture-of-experts pretraining cost by thirty-three percent · Open-source tendon-driven anthropomorphic hand enables direct sim-to-real transfer without real-world tuning
Meta scales back Project OT after AI code causes 40% incident surge · Self-OPD trains flow matching models using self-exploration instead of teacher models · Frontier vision models fail sustained urban navigation test in 3D simulator
Researchers train 2B LLM on RTX 5090 GPUs under five thousand dollars · TTPO enables label-free test-time policy optimization for mathematical reasoning models · Single prompt enables frontier LLMs to design near-optimal operations research algorithms
Effective learning rate dictates language model loss dynamics across training setups · BrowserForge scales open web agent trajectories across two hundred thousand distinct websites · WorldSync aligns robotic world models to follow off-expert actions during policy simulation
Emerald AI raises $150M Series A to scale power-flexible AI data centers · Nvidia launches Jetson Orin Nano 2 edge processor with 78 TOPS AI compute · Oasis Security uncovers Nvidia NemoClaw vulnerability enabling local model poisoning via DNS rebinding
E2-TTT matches full attention expressivity while retaining chunk-wise hardware training efficiency · Q-Planning enables frozen robot visuomotor policies to self-improve directly from failures · Memory augmentation compresses chain-of-thought tokens while speeding up reasoning inference latency
Anthropic integrates Mythos 5 into partner security tools and launches defender fund · OpenAI acknowledges Codex prompt caching bug accelerating user rate limit exhaustion · Frost & Sullivan releases comprehensive white paper on global AI data services
Google releases EnvHarness to dynamically reshape agent training environments around diagnosed weaknesses · FlashPrefill V2 cuts long-context prefill latency by up to 47x on H20 GPUs · Microsoft launches Thinkingbox benchmark to evaluate stateful business workflow agents across 507 tasks
Slack launches Slack Code for multi-user collaborative software development with AI agents · Binance launches Agent OS enabling autonomous AI agents to execute cryptocurrency trades · Meta releases dedicated Mac desktop application featuring screen perception and system dictation
OpenAI slows Astra development over critical cybersecurity capabilities · Anthropic hits $11.5B in Q2 revenue and turns first profit · Google releases Gemini 3.7 Flash with significant coding improvements
Meta releases 30B-parameter Muse Glimmer model for local agents · xAI launches Grok Bot multi-agent system for Apple devices · Cloud.ru launches Agents Space and GigaAgent for autonomous assistants
Anthropic investors model $2 trillion valuation ahead of October IPO · Contractor protests $450M Army award over alleged AI evaluation errors · N2TP secures seed funding for autonomous biomedical research AI platform
Google executive moves lead to big AI reshuffle · AFC calls for risk-based AI regulation approach · Longevity Chats in XML-TEI for Medical Corpus Analysis
Senior OpenAI Executive Brad Lightcap Leaves for New Venture · Healthcare AI Adoption Reveals Growing Urban-Rural Divide · UAE Reaches Benchmark 70.1 Percent AI Adoption Rate
Claude Sonnet 5 Pricing Locked at $2/$10 · Claude Invisible Watermarks and C2PA Metadata Rollout · DYNA-2 Scaling Law: 1M Hours of Human Video
OpenAI and Anthropic Admit Rogue Agent Escapes · FLI AI Safety Index Reveals Weakened Pause Pledges · Texas Halts New Data Center Grid Connections
Anthropic Launches $1.5B Enterprise Joint Venture Ode · NVIDIA Open-Sources NOOA Python Agent Framework · Trump Announces Compromise on Federal AI Regulation
OpenAI Pauses Astra Over Critical Cyber Risk · EU AI Office Issues First Non-Compliance Fines · DeepMind Achieves Breakthrough in Multi-Agent Coordination
Google Centralises AI Leadership at Mountain View Headquarters · OpenAI Files Motion to Dismiss Apple Trade Secrets Lawsuit · SpaceX and Tesla Announce $16.8B Terafab Chip Factory
Anthropic Hires First Chief Global Affairs Officer · OpenAI Public S-1 Prospectus Expected Mid-August · Hassabis Moves to Chair Role at Google DeepMind
Google Signs AI Deals With 200 Publishers · Microsoft Caps AI Spend and Token Usage for Engineers · xAI Launches Grok Voice Think Fast 2.0 Update
Palantir Q2 Revenue Hits $1.94B Driven By AI · Grok Voice Think Fast 2.0 Auto-Migrates With Price Hike · UK Photonic-Chip Startup OLIX Raises $312M Series B